CN111612213B - Section constraint intelligent early warning method and system based on deep learning - Google Patents
Section constraint intelligent early warning method and system based on deep learning Download PDFInfo
- Publication number
- CN111612213B CN111612213B CN202010284014.6A CN202010284014A CN111612213B CN 111612213 B CN111612213 B CN 111612213B CN 202010284014 A CN202010284014 A CN 202010284014A CN 111612213 B CN111612213 B CN 111612213B
- Authority
- CN
- China
- Prior art keywords
- data
- constraint
- section
- power grid
- parameters
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 33
- 238000013135 deep learning Methods 0.000 title claims abstract description 18
- 238000010200 validation analysis Methods 0.000 claims abstract description 21
- 239000013598 vector Substances 0.000 claims description 40
- 238000004422 calculation algorithm Methods 0.000 claims description 20
- 238000013528 artificial neural network Methods 0.000 claims description 17
- 238000005192 partition Methods 0.000 claims description 13
- 238000012549 training Methods 0.000 claims description 12
- 238000012216 screening Methods 0.000 claims description 8
- 238000010801 machine learning Methods 0.000 claims description 7
- 230000006870 function Effects 0.000 claims description 5
- 238000003064 k means clustering Methods 0.000 claims description 5
- 239000011159 matrix material Substances 0.000 claims description 5
- 238000010977 unit operation Methods 0.000 claims description 5
- 230000004913 activation Effects 0.000 claims description 4
- 238000004364 calculation method Methods 0.000 claims description 4
- 238000010276 construction Methods 0.000 claims description 3
- 238000000605 extraction Methods 0.000 claims description 3
- 238000003062 neural network model Methods 0.000 claims description 3
- 238000012545 processing Methods 0.000 claims description 3
- 230000001537 neural effect Effects 0.000 claims description 2
- 239000000284 extract Substances 0.000 abstract description 5
- 238000010586 diagram Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004590 computer program Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000008520 organization Effects 0.000 description 1
- 238000010248 power generation Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/06—Energy or water supply
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
- Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
- Y04S10/00—Systems supporting electrical power generation, transmission or distribution
- Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Business, Economics & Management (AREA)
- Economics (AREA)
- Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Human Resources & Organizations (AREA)
- Strategic Management (AREA)
- Marketing (AREA)
- Computing Systems (AREA)
- Software Systems (AREA)
- General Business, Economics & Management (AREA)
- Mathematical Physics (AREA)
- General Engineering & Computer Science (AREA)
- Tourism & Hospitality (AREA)
- Molecular Biology (AREA)
- Evolutionary Computation (AREA)
- Data Mining & Analysis (AREA)
- Computational Linguistics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Primary Health Care (AREA)
- Development Economics (AREA)
- Water Supply & Treatment (AREA)
- Public Health (AREA)
- Game Theory and Decision Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Supply And Distribution Of Alternating Current (AREA)
Abstract
The invention discloses a section constraint intelligent early warning method and a section constraint intelligent early warning system based on deep learning, wherein the method extracts power grid operation data with highest similarity with power grid operation data to be predicted from K-class typical operation modes to obtain first section constraint data; inputting the operation parameters of the power grid to be predicted into a section constraint prediction model to obtain second section constraint data and constraint validation probability; extracting the same constraint parameters in the first section constraint data and the second section constraint data, and removing constraint parameters with constraint effective probability not larger than a first threshold value to obtain first early warning data; extracting constraint parameters which are different from the constraint parameters of the first section from the constraint data of the second section, and removing constraint parameters with the constraint validation probability not larger than a second threshold value to obtain second early warning data; obtaining an early warning result from the first early warning data and the second early warning data; the method can improve the accuracy of the section constraint early warning result, thereby improving the operation safety of the power grid and the spot market clearing efficiency.
Description
Technical Field
The invention relates to the field of power systems, in particular to a section constraint intelligent early warning method and system based on deep learning.
Background
The effective early warning of section constraint can provide useful reference information for operators, and ensure that the spot market clear result meets the requirement of the power grid safe operation constraint. In addition, the effective early warning of the section constraint can reflect the influence of massive safety constraints on the operation safety of the system, further help a dispatcher to reduce redundant constraints in the market clearing model, improve the model solving efficiency and greatly reduce the calculation burden of the market clearing model.
In the prior art, the section constraint mainly sets deterministic scenes through manual experience, lacks certain scientific basis, and especially determines the power generation plan of a marketized part of the unit according to market transaction results under the condition of daily spot market transaction organization, so that the number of possible scenes is increased sharply, and the situations of missing alarm, false alarm and the like can be caused only through a manual setting mode, thereby influencing the operation safety of a power grid and the spot market clearing efficiency.
Disclosure of Invention
The embodiment of the invention provides a section constraint intelligent early warning method and a section constraint intelligent early warning system based on deep learning, which can improve the accuracy of section constraint early warning results, thereby improving the operation safety of a power grid and the spot market clearing efficiency.
In order to solve the technical problems, an embodiment of the present invention provides a section constraint intelligent early warning method based on deep learning, including:
acquiring power grid operation data to be predicted and power grid operation parameters to be predicted;
extracting power grid operation data with highest similarity with the power grid operation data to be predicted from K-class typical operation modes, and obtaining first section constraint data;
inputting the operation parameters of the power grid to be predicted into a section constraint prediction model to obtain second section constraint data and constraint effective probabilities corresponding to constraint parameters in the second section constraint data;
extracting constraint parameters which are the same as the constraint parameters of the first section from the constraint data of the second section, and removing constraint parameters with the constraint validation probability not larger than a first preset threshold value to obtain first early warning data;
extracting constraint parameters which are different from the first section constraint data from the second section constraint data, and removing constraint parameters with the effective probability not larger than a second preset threshold value to obtain second early warning data;
and obtaining an early warning result according to the first early warning data and the second early warning data.
As a preferable scheme, the power grid operation data to be predicted comprises hydroelectric generating set operation data, wind generating set operation data, photovoltaic generating set operation data, line operation/operation-withdrawal state data, main transformer operation/operation-withdrawal state data and partition load data; wherein,
the operation data of the hydroelectric generating set comprises an output value of the hydroelectric generating set and the number of the hydroelectric generating set;
the running data of the wind turbine comprises a wind turbine output value and the number of wind turbines;
the photovoltaic unit operation data comprise a photovoltaic unit output value and the number of photovoltaic units;
the line operation/return operation state data comprises a state value of line operation/return operation and the number of main network lines;
the main transformer operation/return operation state data comprises main transformer operation/return operation state values and the number of main transformers;
the partition load data comprises a load value and a load number.
As a preferred scheme, the K-class typical operation mode is obtained by a clustering algorithm, and comprises the following specific steps:
based on an AP clustering algorithm or a K-means clustering algorithm, clustering processing is carried out on the power grid operation data, and a K-class typical operation mode is obtained.
As a preferred solution, the extracting, from the K-class typical operation modes, the power grid operation data with the highest similarity to the power grid operation data to be predicted, and obtaining first section constraint data specifically includes:
calculating the similarity between the characteristic vector of the power grid operation data to be predicted and the characteristic vector of the power grid operation data in each of K types of typical operation modes based on a dynamic time warping algorithm or a cosine similarity algorithm or an information entropy method, extracting the power grid operation data in the typical operation mode with the highest similarity, and obtaining the power grid operation data to be processed;
and extracting the power grid operation data with the section constraint from the power grid operation data to be processed, and obtaining first section constraint data.
Preferably, the step of calculating the feature vector is as follows:
wherein ,Wi Is a feature vector of the grid operation data,characteristic vector of operation data of hydroelectric generating set, +.>Characteristic vectors of the operating data of the wind turbine generator system, +.>Characteristic direction of operation data of photovoltaic unitQuantity (S)>Feature vector for line commissioning/decommissioning status data,/->Feature vector for main transformer delivery/return state data, +.>For the eigenvector of the partition load data, P GH，j,t Output value of hydroelectric generating set j in t period, N W Number of hydroelectric generating sets, P GW，t Output value, P of whole-network wind turbine generator system in t period GS，t Output value of full-network photovoltaic unit in t period, alpha L，t Status value of commissioning/decommissioning of line i, N L Alpha is the number of main network lines T，t Operational/return status value of main transformer i, N T The number of main transformers, P LD，t Load value, N, in partition i LD The number of loads.
Preferably, the power grid operation parameters to be predicted include: wind speed, photovoltaic machine output value, power plant quotation and load.
As a preferable scheme, the construction method of the section constraint prediction model comprises the following steps:
and taking the power grid operation parameters and the section constraint results corresponding to the power grid operation parameters as training sets, and inputting the training sets into a machine learning model for training so as to construct a trained section constraint prediction module.
As a preferred scheme, the machine learning model is obtained by a deep neural network model, specifically:
Y＝f n (…f 2 (W 2 f 1 (W 1 X+b 1 ))+b 2 )
wherein f is an activation function, n is the number of layers of the neural network, and the weight W i Is d i+1 ×d i Matrix of dimensions, offset b i Is d i+1 Vector of dimensions d i Is the deep neural networkThe number of neural networks in the i layer.
Correspondingly, the invention also provides a section constraint intelligent early warning system based on deep learning, which comprises:
the acquisition module is used for acquiring the power grid operation data to be predicted and the power grid operation parameters to be predicted;
the extraction module is used for extracting the power grid operation data with the highest similarity with the power grid operation data to be predicted from K-class typical operation modes to obtain first section constraint data;
the prediction module is used for inputting the operation parameters of the power grid to be predicted into a section constraint prediction model to obtain second section constraint data and constraint effective probabilities corresponding to constraint parameters in the second section constraint data;
the first screening module is used for extracting constraint parameters which are the same as the constraint parameters of the first section from the constraint data of the second section, and eliminating constraint parameters with the effective probability of constraint not more than a first preset threshold value to obtain first early warning data;
the second screening module is used for extracting constraint parameters which are different from the first section constraint data from the second section constraint data, and eliminating constraint validation probability not larger than a second preset threshold constraint parameter to obtain second early warning data;
and the analysis module is used for obtaining an early warning result according to the first early warning data and the second early warning data.
The embodiment of the invention has the following beneficial effects:
according to the section constraint intelligent early warning method based on deep learning, firstly, power grid operation data with highest similarity with power grid operation data to be predicted is extracted from K-class typical operation modes, and first section constraint data is obtained; inputting the operation parameters of the power grid to be predicted into a section constraint prediction model to obtain second section constraint data and constraint effective probabilities corresponding to the constraint parameters in the second section constraint data; extracting the same constraint parameters in the first section constraint data and the second section constraint data, and removing constraint parameters with constraint effective probability not larger than a first threshold value to obtain first early warning data; extracting constraint parameters which are different from the constraint parameters of the first section from the constraint data of the second section, and removing constraint parameters with the constraint validation probability not larger than a second threshold value to obtain second early warning data; compared with the prior art that the section constraint is obtained through manual experience, the method does not need to rely on manual experience to obtain the section constraint, and accuracy is high; in addition, the method acquires the same constraint parameters from the first section constraint data and the second section constraint, eliminates constraint parameters with constraint validation probability not larger than a first preset threshold, improves accuracy of a section constraint result, extracts constraint parameters different from the first section constraint from the second section constraint data, eliminates constraint parameters with constraint validation probability lower than a second preset threshold, further improves accuracy of the section constraint result, and further improves operation safety of a power grid and spot market clearing efficiency.
Drawings
FIG. 1 is a schematic flow chart of a first embodiment of a section constraint intelligent early warning method based on deep learning;
fig. 2 is a schematic structural diagram of a second embodiment of a section constraint intelligent early warning system based on deep learning.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
First embodiment:
referring to fig. 1, a flow chart of an embodiment of a section constraint intelligent early warning method based on deep learning is provided. Referring to fig. 1, the method includes steps 101 to 106, which are specifically as follows:
step 101: and acquiring the power grid operation data to be predicted and the power grid operation parameters to be predicted.
In one preferred embodiment, the power grid operation data to be predicted comprises hydroelectric generating set operation data, wind generating set operation data, photovoltaic generating set operation data, line commissioning/outage state data, main transformer commissioning/outage state data and partition load data; the operation data of the hydroelectric generating set comprises an output value of the hydroelectric generating set and the number of the hydroelectric generating set; the wind turbine running data comprise wind turbine output values and the number of wind turbines; the photovoltaic unit operation data comprise a photovoltaic unit output value and the number of photovoltaic units; the line operation/return operation state data comprises a state value of line operation/return operation and the number of main network lines; the main transformer operation/return operation state data comprises main transformer operation/return operation state values and the number of main transformers; the partition load data comprises load values and load numbers; the power grid operation parameters to be predicted comprise: wind speed, photovoltaic machine output value, power plant quotation and load.
In the embodiment, as new energy sources such as wind, light and load have randomness, and the quotation of the power plant has uncertainty due to the influence of climate conditions, the wind, light, load and quotation of the power plant all have certain influence on the operation safety of the power system. According to the method, the influence of the dimensions on the section constraint is reduced from the dimensions of wind, light, load and power plant quotation, so that the accuracy of the section constraint prediction result is improved, and the operation safety of the power system is further improved.
Step 102: and extracting the power grid operation data with highest similarity with the power grid operation data to be predicted from the K-class typical operation modes, and obtaining first section constraint data.
In one preferred embodiment, the K-class typical operation mode is obtained by a clustering algorithm, and the specific steps are as follows: based on an AP clustering algorithm or a K-means clustering algorithm, clustering processing is carried out on the power grid operation data of the historical samples, and a K-class typical operation mode is obtained.
In this embodiment, the K-class typical operation mode is not defined as an operation mode of a corresponding power system, but is a clustering-capable class number with similar power grid operation characteristics, the optimal clustering number K is automatically calculated through an AP clustering algorithm or a K-means clustering algorithm, and different operation modes of the K-class are obtained by clustering according to power grid operation data of a historical sample.
In one preferred embodiment, step 102 is specifically: calculating the similarity between the characteristic vector of the power grid operation data to be predicted and the characteristic vector of the power grid operation data in each typical operation mode in the K-type typical operation modes based on a dynamic time warping algorithm or a cosine similarity algorithm or an information entropy method, extracting the power grid operation data in the typical operation mode with the highest similarity, and obtaining the power grid operation data to be processed; extracting power grid operation data with section constraint from power grid operation data to be processed, and obtaining first section constraint data; the feature vector is calculated as follows:
wherein ,Wi Is a feature vector of the grid operation data,characteristic vector of operation data of hydroelectric generating set, +.>Characteristic vectors of the operating data of the wind turbine generator system, +.>Feature vector of photovoltaic unit operation data +.>Feature vector for line commissioning/decommissioning status data,/->Feature vector for main transformer delivery/return state data, +.>For the eigenvector of the partition load data, P GH，j,t Output value of hydroelectric generating set j in t period, N W Number of hydroelectric generating sets, P GW，t Output value, P of whole-network wind turbine generator system in t period GS，t Output value of full-network photovoltaic unit in t period, alpha L，t Status value of commissioning/decommissioning of line i, N L Alpha is the number of main network lines T，t Operational/return status value of main transformer i, N T The number of main transformers, P LD，t Load value, N, in partition i LD The number of loads.
For example, a first eigenvector W of the grid operation data to be predicted is obtained from the calculation formula of the eigenvector j The method comprises the steps of carrying out a first treatment on the surface of the Obtaining a second characteristic vector W of the power grid operation data under the i-th type typical operation mode according to a calculation formula of the characteristic vector i Wherein K is more than or equal to i is more than or equal to 1; calculating information entropy of the first feature vector and the second feature vector based on an information entropy value method, if the absolute value of a difference value of the information entropy is smaller than or equal to 5%, the power grid operation data to be predicted belong to an i-th type typical mode, and extracting power grid operation data with section constraint in the power grid operation data in the i-th type typical mode to obtain first section constraint data; the information entropy method comprises the following steps:
wherein the second feature vector W i Corresponding class i canonicalThe historical power grid operation data is divided into 10 sections according to the upper limit value and the lower limit value of the data per se in an operation mode; p is p k For the statistical probability value that the related power grid operation data falls into the above-defined interval, the related power grid operation data can be taken as a first feature vector W j Grid operation data or second feature vector W i And the power grid operation data correspondingly generate information entropy of the first characteristic vector and the second characteristic vector.
In this embodiment, since the K-type typical operation mode is obtained from historical grid operation data, and the effective section of the historical grid operation data is also determined, the section-constrained grid operation data in the i-th typical mode can be directly extracted.
Step 103: and inputting the operation parameters of the power grid to be predicted into a section constraint prediction model to obtain second section constraint data and constraint effective probabilities corresponding to the constraint parameters in the second section constraint data.
In one preferred embodiment, the method for constructing the section constraint prediction model is as follows: and taking the power grid operation parameters and the section constraint results corresponding to the power grid operation parameters as training sets, and inputting the training sets into a machine learning model for training so as to construct a trained section constraint prediction module.
In one preferred embodiment, the machine learning model is obtained from a deep neural network model, specifically:
Y＝f n (…f 2 (W 2 f 1 (W 1 X+b 1 ))+b 2 )
wherein f is an activation function, n is the number of layers of the neural network, and the weight W i Is d i+1 ×d i Matrix of dimensions, offset b i Is d i+1 Vector of dimensions d i Is the number of the neural networks of the ith layer of the deep neural network.
Step 104: and after extracting constraint parameters which are the same as the constraint parameters of the first section from the constraint data of the second section, eliminating constraint parameters of which the constraint effective probability is not more than a first preset threshold value, and obtaining first early warning data.
In this embodiment, after extracting constraint parameters identical to those of the first section constraint data from the second section constraint data, eliminating constraint parameters with constraint validation probability not greater than 50% to obtain first early warning data.
In the embodiment, the same constraint parameters are obtained from the first section constraint data and the second section constraint, constraint parameters with constraint validation probability not larger than a first preset threshold are removed, and accuracy of section constraint results is improved, so that operation safety of a power grid and spot market clearing efficiency are improved.
Step 105: and after extracting constraint parameters which are different from the constraint parameters of the first section from the constraint parameters of the second section, eliminating constraint validation probability not larger than a constraint parameter of a second preset threshold value, and obtaining second early warning data.
In this embodiment, after constraint parameters different from the first section constraint data are extracted from the second section constraint data, constraint parameters with constraint validation probability not greater than 90% are removed, so as to obtain second early warning data.
In this embodiment, constraint parameters different from those in the first section constraint are extracted from the second section constraint data, and constraint parameters with constraint validation probability lower than a second preset threshold are removed, so that accuracy of a section constraint result is further improved, and operation safety of a power grid and spot market clearing efficiency are further improved.
Step 106: and obtaining an early warning result according to the first early warning data and the second early warning data.
From the above, the method provided by the embodiment of the invention extracts the power grid operation data with highest similarity with the power grid operation data to be predicted from K-type typical operation modes to obtain the first section constraint data; inputting the operation parameters of the power grid to be predicted into a section constraint prediction model to obtain second section constraint data and constraint effective probabilities corresponding to the constraint parameters in the second section constraint data; extracting the same constraint parameters in the first section constraint data and the second section constraint data, and removing constraint parameters with constraint effective probability not larger than a first threshold value to obtain first early warning data; extracting constraint parameters which are different from the constraint parameters of the first section from the constraint data of the second section, and removing constraint parameters with the constraint validation probability not larger than a second threshold value to obtain second early warning data; compared with the prior art that the section constraint is obtained through manual experience, the method does not need to rely on manual experience to obtain the section constraint, and accuracy is high; in addition, the method acquires the same constraint parameters from the first section constraint data and the second section constraint, eliminates constraint parameters with constraint validation probability not larger than a first preset threshold, improves accuracy of a section constraint result, extracts constraint parameters different from the first section constraint from the second section constraint data, eliminates constraint parameters with constraint validation probability lower than a second preset threshold, further improves accuracy of the section constraint result, and further improves operation safety of a power grid and spot market clearing efficiency.
Second embodiment
Referring to fig. 2, the present invention provides a structural schematic diagram of a second embodiment of a section constraint intelligent early warning system based on deep learning, the system includes: the system comprises an acquisition module 201, an extraction module 202, a prediction module 203, a first screening module 204, a second screening module 205 and an analysis module 206.
An obtaining module 201, configured to obtain power grid operation data to be predicted and power grid operation parameters to be predicted;
the extracting module 202 is configured to extract, from the K-class typical operation modes, power grid operation data with highest similarity to power grid operation data to be predicted, and obtain first section constraint data;
the prediction module 203 is configured to input the power grid operation parameter to be predicted into a section constraint prediction model, and obtain second section constraint data and constraint effective probabilities corresponding to constraint parameters in the second section constraint data;
the first screening module 204 is configured to extract constraint parameters that are the same as the constraint parameters of the first section from the second section constraint data, and then reject constraint parameters with a constraint validation probability not greater than a first preset threshold, so as to obtain first early warning data;
the second screening module 205 is configured to extract constraint parameters that are different from the first section constraint data from the second section constraint data, and reject constraint parameters that have a constraint validation probability not greater than a second preset threshold value, so as to obtain second early warning data;
and the analysis module 206 is configured to obtain an early warning result according to the first early warning data and the second early warning data.
The more detailed working principle and flow of the embodiment can be, but not limited to, a section constraint intelligent early warning method based on deep learning of the first embodiment.
From the above, the invention does not need to rely on manual experience to obtain section constraint, and has high accuracy; in addition, the method acquires the same constraint parameters from the first section constraint data and the second section constraint, eliminates constraint parameters with constraint validation probability not larger than a first preset threshold, improves accuracy of a section constraint result, extracts constraint parameters different from the first section constraint from the second section constraint data, eliminates constraint parameters with constraint validation probability lower than a second preset threshold, further improves accuracy of the section constraint result, and further improves operation safety of a power grid and spot market clearing efficiency.
Those skilled in the art will appreciate that implementing all or part of the above-described methods in accordance with the embodiments may be accomplished by way of a computer program stored on a computer readable storage medium, which when executed may comprise the steps of the embodiments of the methods described above. The storage medium may be a magnetic disk, an optical disk, a Read-Only Memory (ROM), a random access Memory (Random Access Memory, RAM), or the like.
While the foregoing is directed to the preferred embodiments of the present invention, it will be appreciated by those skilled in the art that changes and modifications may be made without departing from the principles of the invention, such changes and modifications are also intended to be within the scope of the invention.
Claims (5)
1. A section constraint intelligent early warning method based on deep learning is characterized by comprising the following steps:
obtaining power grid operation data to be predicted and power grid operation parameters to be predicted, wherein the power grid operation parameters to be predicted comprise: wind speed, photovoltaic machine output value, power plant quotation and load; the power grid operation data to be predicted comprises hydroelectric generating set operation data, wind generating set operation data, photovoltaic generating set operation data, line operation/operation withdrawal state data, main transformer operation/operation withdrawal state data and partition load data; the operation data of the hydroelectric generating set comprises an output value of the hydroelectric generating set and the number of the hydroelectric generating set; the running data of the wind turbine comprises a wind turbine output value and the number of wind turbines; the photovoltaic unit operation data comprise a photovoltaic unit output value and the number of photovoltaic units; the line operation/return operation state data comprises a state value of line operation/return operation and the number of main network lines; the main transformer operation/return operation state data comprises main transformer operation/return operation state values and the number of main transformers; the partition load data comprises a load value and a load number;
extracting power grid operation data with highest similarity with the power grid operation data to be predicted from K-class typical operation modes to obtain first section constraint data, wherein the K-class typical operation modes are obtained by a clustering algorithm, and clustering the power grid operation data based on an AP (access point) clustering algorithm or a K-means clustering algorithm to obtain K-class typical operation modes;
inputting the operation parameters of the power grid to be predicted into a section constraint prediction model to obtain second section constraint data and constraint effective probabilities corresponding to all constraint parameters in the second section constraint data, wherein the construction method of the section constraint prediction model comprises the following steps: taking the power grid operation parameters and the section constraint results corresponding to the power grid operation parameters as training sets, and inputting the training sets into a machine learning model for training to construct a trained section constraint prediction module;
extracting constraint parameters which are the same as the constraint parameters of the first section from the constraint data of the second section, and removing constraint parameters with the constraint validation probability not larger than a first preset threshold value to obtain first early warning data;
extracting constraint parameters which are different from the first section constraint data from the second section constraint data, and removing constraint parameters with the effective probability not larger than a second preset threshold value to obtain second early warning data;
and obtaining an early warning result according to the first early warning data and the second early warning data.
2. The intelligent section constraint early warning method based on deep learning of claim 1, wherein the extracting the grid operation data with the highest similarity with the grid operation data to be predicted from the K-type typical operation modes, to obtain the first section constraint data, specifically comprises:
calculating the similarity between the characteristic vector of the power grid operation data to be predicted and the characteristic vector of the power grid operation data in each of K types of typical operation modes based on a dynamic time warping algorithm or a cosine similarity algorithm or an information entropy method, extracting the power grid operation data in the typical operation mode with the highest similarity, and obtaining the power grid operation data to be processed;
and extracting the power grid operation data with the section constraint from the power grid operation data to be processed, and obtaining first section constraint data.
3. The intelligent section constraint early warning method based on deep learning as claimed in claim 2, wherein the feature vector calculation step is as follows:
wherein ,for the feature vector of the network operating data, +.>Characteristic vector of operation data of hydroelectric generating set, +.>Characteristic vectors of the operating data of the wind turbine generator system, +.>Characteristic vector for operating data of photovoltaic unit, < >>Feature vector for line commissioning/decommissioning status data,/->Feature vector for main transformer delivery/return state data, +.>For the eigenvector of the partition load data, +.>For hydroelectric generating setsjAt the position oftForce value of time period->For the number of hydroelectric generating sets, < >>Is in a whole-network wind turbine generator systemtForce value of time period->Is a full-network photovoltaic unittForce value of time period->Is on line withtStatus value of put/put-back in period, +.>For the number of main network lines, < > and->Is mainly varied attTime period commissioning/retirement status value, +.>The number of main transformers>Is in the subareatLoad value of time period>The number of loads.
4. The cross section constraint intelligent early warning method based on deep learning according to claim 1, wherein the machine learning model is obtained by a deep neural network model, specifically:
wherein ,fin order to activate the function,nfor the number of layers of the neural network, weightIs +.>Matrix of dimensions, bias->Is oneVector of dimension,/->The number of neural networks of the i-th layer of the deep neural network, < ->For the corresponding activation function of the layer 2 neural network, < ->For the corresponding activation function of the layer 1 neural network, < ->Is +.>Matrix of dimensions->The number of neural networks of layer 3 of the deep neural network, < ->The number of the neural networks of the layer 2 of the deep neural network, < ->Is +.>Matrix of dimensions->The number of the neural networks of the layer 1 of the deep neural network, < ->Is +.>Vector of dimension,/->Is +.>Vector of dimension,/->Is the deep neural network->The number of the neural networks of the layers, X is the inputted section constraint result.
5. Section constraint intelligent early warning system based on deep learning, characterized by comprising:
the system comprises an acquisition module, a prediction module and a prediction module, wherein the acquisition module is used for acquiring power grid operation data to be predicted and power grid operation parameters to be predicted, and the power grid operation parameters to be predicted comprise: wind speed, photovoltaic machine output value, power plant quotation and load; the power grid operation data to be predicted comprises hydroelectric generating set operation data, wind generating set operation data, photovoltaic generating set operation data, line operation/operation withdrawal state data, main transformer operation/operation withdrawal state data and partition load data; the operation data of the hydroelectric generating set comprises an output value of the hydroelectric generating set and the number of the hydroelectric generating set; the running data of the wind turbine comprises a wind turbine output value and the number of wind turbines; the photovoltaic unit operation data comprise a photovoltaic unit output value and the number of photovoltaic units; the line operation/return operation state data comprises a state value of line operation/return operation and the number of main network lines; the main transformer operation/return operation state data comprises main transformer operation/return operation state values and the number of main transformers; the partition load data comprises a load value and a load number;
the extraction module is used for extracting the power grid operation data with the highest similarity with the power grid operation data to be predicted from K-class typical operation modes to obtain first section constraint data, wherein the K-class typical operation modes are obtained by a clustering algorithm, and clustering processing is carried out on the power grid operation data based on an AP clustering algorithm or a K-means clustering algorithm to obtain K-class typical operation modes;
the prediction module is used for inputting the operation parameters of the power grid to be predicted into a section constraint prediction model to obtain second section constraint data and constraint effective probabilities corresponding to all constraint parameters in the second section constraint data, wherein the construction method of the section constraint prediction model comprises the following steps: taking the power grid operation parameters and the section constraint results corresponding to the power grid operation parameters as training sets, and inputting the training sets into a machine learning model for training to construct a trained section constraint prediction module;
the first screening module is used for extracting constraint parameters which are the same as the constraint parameters of the first section from the constraint data of the second section, and eliminating constraint parameters with the effective probability of constraint not more than a first preset threshold value to obtain first early warning data;
the second screening module is used for extracting constraint parameters which are different from the first section constraint data from the second section constraint data, and eliminating constraint validation probability not larger than a second preset threshold constraint parameter to obtain second early warning data;
and the analysis module is used for obtaining an early warning result according to the first early warning data and the second early warning data.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010284014.6A CN111612213B (en) | 2020-04-10 | 2020-04-10 | Section constraint intelligent early warning method and system based on deep learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010284014.6A CN111612213B (en) | 2020-04-10 | 2020-04-10 | Section constraint intelligent early warning method and system based on deep learning |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111612213A CN111612213A (en) | 2020-09-01 |
| CN111612213B true CN111612213B (en) | 2023-10-10 |
Family
ID=72197762
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010284014.6A Active CN111612213B (en) | 2020-04-10 | 2020-04-10 | Section constraint intelligent early warning method and system based on deep learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111612213B (en) |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2013084136A (en) * | 2011-10-11 | 2013-05-09 | Nippon Hoso Kyokai <Nhk> | Semantic analysis device and program thereof |
| CN108009673A (en) * | 2017-11-24 | 2018-05-08 | 国网北京市电力公司 | Novel load Forecasting Methodology and device based on deep learning |
| CN108183481A (en) * | 2018-01-29 | 2018-06-19 | 中国电力科学研究院有限公司 | One kind quickly sentences steady method and system based on deep learning power grid |
| CN108985515A (en) * | 2018-07-24 | 2018-12-11 | 国网河南省电力公司电力科学研究院 | A kind of new energy based on independent loops neural network goes out force prediction method and system |
| CN109784692A (en) * | 2018-12-29 | 2019-05-21 | 重庆大学 | A kind of fast and safely constraint economic load dispatching method based on deep learning |
| CN109995031A (en) * | 2019-05-05 | 2019-07-09 | 重庆大学 | Probabilistic Load Flow deep learning calculation method based on physical model |
-
2020
- 2020-04-10 CN CN202010284014.6A patent/CN111612213B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2013084136A (en) * | 2011-10-11 | 2013-05-09 | Nippon Hoso Kyokai <Nhk> | Semantic analysis device and program thereof |
| CN108009673A (en) * | 2017-11-24 | 2018-05-08 | 国网北京市电力公司 | Novel load Forecasting Methodology and device based on deep learning |
| CN108183481A (en) * | 2018-01-29 | 2018-06-19 | 中国电力科学研究院有限公司 | One kind quickly sentences steady method and system based on deep learning power grid |
| CN108985515A (en) * | 2018-07-24 | 2018-12-11 | 国网河南省电力公司电力科学研究院 | A kind of new energy based on independent loops neural network goes out force prediction method and system |
| CN109784692A (en) * | 2018-12-29 | 2019-05-21 | 重庆大学 | A kind of fast and safely constraint economic load dispatching method based on deep learning |
| CN109995031A (en) * | 2019-05-05 | 2019-07-09 | 重庆大学 | Probabilistic Load Flow deep learning calculation method based on physical model |
Non-Patent Citations (1)
| Title |
|---|
| 电力现货市场仿真系统的设计与实现;江泽志;梁立全;刘滨;陈皓勇;伍景阳;;浙江电力(01);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111612213A (en) | 2020-09-01 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Zhu et al. | Hierarchical deep learning machine for power system online transient stability prediction | |
| Ghaderi et al. | Deep forecast: Deep learning-based spatio-temporal forecasting | |
| Blanco et al. | An efficient robust solution to the two-stage stochastic unit commitment problem | |
| CN109447441B (en) | Transient stability risk assessment method considering uncertainty of new energy unit | |
| CN110365041B (en) | Wind power multi-scene robust scheduling method based on gan scene simulation | |
| CN104077632A (en) | Wind power field power prediction method based on deep neural network | |
| CN109038660A (en) | A kind of wind-electricity integration System Reactive Power planing method considering quiet Enhancement of Transient Voltage Stability | |
| CN103258142A (en) | Wind power plant short-term wind speed forecasting method based on weather patterns | |
| CN111832812A (en) | Wind power short-term prediction method based on deep learning | |
| CN112632288A (en) | Power dispatching system and method based on knowledge graph | |
| CN103049609B (en) | Wind power multi-stage scene simulation method | |
| CN113344288B (en) | Cascade hydropower station group water level prediction method and device and computer readable storage medium | |
| CN105140967B (en) | A kind of appraisal procedure of the demand of peak regulation containing New-energy power system | |
| Zhu et al. | Short term forecast of wind power generation based on SVM with pattern matching | |
| Hou et al. | Spatial distribution assessment of power outage under typhoon disasters | |
| Yan et al. | Deep learning based total transfer capability calculation model | |
| CN113627685B (en) | Wind driven generator power prediction method considering wind power internet load limit | |
| CN111612213B (en) | Section constraint intelligent early warning method and system based on deep learning | |
| CN115986945B (en) | Electric power equipment monitoring method, equipment and medium based on industrial Internet | |
| Guo et al. | Wind speed forecasting of genetic neural model based on rough set theory | |
| Riahinia et al. | A neural network-based model for wind farm output in probabilistic studies of power systems | |
| CN111026741A (en) | Data cleaning method and device based on time series similarity | |
| Zhenhao et al. | Prediction of wind power ramp events based on deep neural network | |
| CN114358415A (en) | Typhoon season overhead line trip prediction method based on interactive hidden Markov model | |
| Haidar et al. | An intelligent load shedding scheme using neural networks and neuro-fuzzy |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |